CN114519806A - Ocean wave level observation model training method and system - Google Patents
Ocean wave level observation model training method and system Download PDFInfo
- Publication number
- CN114519806A CN114519806A CN202210112584.6A CN202210112584A CN114519806A CN 114519806 A CN114519806 A CN 114519806A CN 202210112584 A CN202210112584 A CN 202210112584A CN 114519806 A CN114519806 A CN 114519806A
- Authority
- CN
- China
- Prior art keywords
- wave
- data
- level observation
- observation model
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The application provides an ocean wave level observation model training method and system, and the method comprises the following steps: obtaining sea wave video, ultra-wide angle sky video and meteorological observation data; performing secondary encapsulation on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain encapsulated sea wave data, encapsulated sky data and encapsulated meteorological data; screening the packaged sea wave data according to the packaged sky data and the packaged meteorological data to obtain effective sea wave data; performing deep learning modeling on effective sea wave data according to a preset GPU convolutional neural network to obtain a first wave-level observation model; and carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model. Therefore, by implementing the implementation mode, a high-quality ocean wave level observation model can be trained, so that ocean waves can be observed with lower cost, higher precision and higher stability.
Description
Technical Field
The application relates to the field of sea wave observation, in particular to a training method and a system for an ocean wave level observation model.
Background
At present, sea wave observation is mostly focused on offshore and coastal, and the continuously updated offshore sea wave observation method can better observe the grade of the offshore sea wave. However, according to the statistics of the wave grades along the offshore area, the proportion of light waves to the following is about 70%, the proportion of medium waves is about 30%, and no high-grade waves exist except for occasional short and big waves influenced by severe weather processes such as typhoons.
However, wave changes in the ocean typically occur at all levels in the wave hierarchy depending on the season and the sea area. Therefore, the offshore wave and ocean wave have great difference, so that the offshore wave observation method is not suitable for observing ocean waves.
Meanwhile, most of the waves observed through the near-shore video are near-shore waves, the wave heights between the near-shore waves and ocean waves are the same, and the wave-shaped near-shore video of the stormy waves and the surge waves has no way of observing or recording the waves. Therefore, the offshore wave and the ocean wave are two different waves, and the energy spectrum configurations generated by the different wave shapes are completely different. It is seen that it is not appropriate to consider ocean waves from offshore waves, which makes it a problem to be solved how ocean waves in the open sea (ocean) should be observed effectively.
Thus, buoy observation, satellite observation, and manual observation have been proposed by those skilled in the art.
However, the buoy observation method is limited by the cost and the application range of the buoy, and a large amount of data cannot be effectively acquired, so that the ocean wave observation effect has extremely high limitation.
In addition, although the satellite observation method can realize large-area wave observation, the lower resolution of the satellite observation method can result in lower precision of a wave model.
As for the manual observation method, the consumed labor cost and physical cost are very high, and the data acquisition depends on the experience of the observer, so that the traceability is weak, and the long-time stable observation cannot be realized.
Disclosure of Invention
An object of the embodiment of the application is to provide a training method and system for ocean wave observation models, which can directly acquire data in a visual form, realize direct observation of ocean waves, and supplement elements such as ocean meteorology, so that a deep learning method can train a high-quality ocean wave observation model for the ocean wave observation models, and thus ocean waves can be observed with lower cost, higher precision and higher stability.
The embodiment of the application provides a training method for an ocean wave-level observation model in a first aspect, which comprises the following steps: obtaining sea wave video, ultra-wide angle sky video and meteorological observation data; the wave video comprises a first long-focus wave video, a second long-focus wave video, a third long-focus wave video and a wide-angle wave video;
performing secondary packaging on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain packaged sea wave data, packaged sky data and packaged meteorological data;
screening the packaged sea wave data according to the packaged sky data and the packaged meteorological data to obtain effective sea wave data;
performing deep learning modeling on the effective sea wave data according to a preset GPU convolutional neural network to obtain a first wave-level observation model; training labels of sea wave grades are set in the effective sea wave data;
and carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
A second aspect of the embodiments of the present application provides an ocean wave-level observation model training system, which includes:
the acquisition unit is used for acquiring sea wave videos, ultra-wide-angle sky videos and meteorological observation data; the wave video comprises a first long-focus wave video, a second long-focus wave video, a third long-focus wave video and a wide-angle wave video;
the packaging unit is used for carrying out secondary packaging on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain packaged sea wave data, packaged sky data and packaged meteorological data;
the screening unit is used for screening the packaged sea wave data according to the packaged sky data and the packaged meteorological data to obtain effective sea wave data;
the modeling unit is used for carrying out deep learning modeling on the effective sea wave data according to a preset GPU convolutional neural network to obtain a first wave-level observation model; training labels of sea wave grades are set in the effective sea wave data;
the modeling unit is further used for carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
A third aspect of embodiments of the present application provides an electronic device, including a memory and a processor, where the memory is used to store a computer program, and the processor runs the computer program to make the electronic device execute the method for training ocean wave-level observation model according to any one of the first aspect of embodiments of the present application.
A fourth aspect of the embodiments of the present application provides a computer-readable storage medium, which stores computer program instructions, where the computer program instructions, when read and executed by a processor, perform the method for training an observation model at ocean wave level according to any one of the first aspect of the embodiments of the present application.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a schematic flow chart of an ocean wave observation model training method according to an embodiment of the present disclosure;
fig. 2 is a schematic structural diagram of an ocean wave observation model training system according to an embodiment of the present disclosure;
fig. 3 is a schematic structural diagram of a shipborne observation camera provided in the embodiment of the present application;
fig. 4 is a real view schematic diagram of ocean waves according to an embodiment of the present application;
fig. 5 is a schematic view of another ocean wave real scene provided in the embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.
Example 1
Referring to fig. 1, fig. 1 is a schematic flow chart of a training method for an ocean wave observation model according to this embodiment. The ocean wave level observation model training method comprises the following steps:
s101, obtaining sea wave video, ultra-wide angle sky video and meteorological observation data; the wave videos comprise a first long-focus wave video, a second long-focus wave video, a third long-focus wave video and a wide-angle wave video.
In this embodiment, the method may acquire the sea wave video through four onboard cameras, and acquire the wide-angle sky video through one onboard super wide-angle camera.
In this embodiment, the four onboard cameras for acquiring the sea wave video include three tele cameras and one wide camera.
In this embodiment, the ultra-wide-angle camera is used for collecting the sky phenomenon directly above the ultra-wide-angle camera, such as sunny days, cloudy days, and the like.
In this embodiment, the method may acquire meteorological observation data through observation at a shipborne automatic meteorological station. Wherein, the meteorological observation data comprise true wind speed, true wind direction, temperature, air pressure, humidity, visibility, weather phenomenon, GPS coordinate position, ship heading, course, ship speed and the like.
Referring to fig. 3, fig. 3 is a schematic structural diagram of a shipborne observation camera. The structure is arranged on a side board at the height of a scientific investigation ship, such as a top of a driving platform, so as to ensure a sufficient visual angle. Wherein the content of the first and second substances,
a1 is a long-focus camera, connected with the controllable tripod head A2, and used for finely observing and recording the fine structure of the sea waves;
a2 is a controllable tripod head, which is connected with the tele camera A1 and the bracket A3, and is used for keeping the tele camera A1 in a horizontal state;
a3 is a bracket, which is connected with the controllable tripod head A2 and the scientific ship body, and is used as a main body bracket of the long-focus camera A1 and the controllable tripod head A2, and provides stable support for the scientific ship;
a4 is a long-focus camera, is connected with a controllable tripod head A5 and is used for finely observing and recording a fine structure of sea waves;
a5 is a controllable tripod head, which is connected with the tele camera A4 and the bracket A6, and is used for keeping the tele camera A4 in a horizontal state;
a6 is a bracket, which is connected with the controllable tripod head A5 and the scientific ship body, and is used as a main body bracket of the long-focus camera A4 and the controllable tripod head A5, and provides stable support for the scientific ship;
a7 is a wide-angle camera, is connected with a controllable tripod head A8 and is used for finely observing and recording a fine structure of sea waves;
a8 is a controllable tripod head, which is connected with the wide-angle camera A7 and the bracket A9, and is used for keeping the camera A7 at the long focus in a horizontal state;
a9 is a bracket, which is connected with the controllable tripod head A8 and the scientific ship body, and is used as a main body bracket of the long-focus camera A7 and the controllable tripod head A8, and provides stable support for the scientific ship;
a10 is a long-focus camera, is connected with a controllable holder A11, is used for finely observing and recording a fine wave structure, exists as a mobile camera, is used for capturing the maximum waves in an area, and simultaneously exists as a supplementary camera, and is used for controlling a worker to acquire data of the concerned area;
a11 is a controllable tripod head, which is connected with the tele camera A10 and the bracket A12, and is used for keeping the tele camera A10 in a horizontal state;
a12 is a bracket, which is connected with the controllable tripod head A11 and the main body of the scientific ship, and is used as a main body bracket for the long-focus camera A10 and the controllable tripod head A11, and provides a stable support for the scientific ship.
Please refer to fig. 4 and 5. Fig. 4 and 5 show a realistic view of ocean waves.
S102, carrying out secondary packaging on the sea wave video, the ultra-wide angle sky video and the meteorological observation data to obtain packaged sea wave data, packaged sky data and packaged meteorological data.
As an optional implementation, the step of performing secondary encapsulation on the sea wave video, the ultra-wide angle sky video, and the meteorological observation data to obtain encapsulated sea wave data, encapsulated sky data, and encapsulated meteorological data includes:
copying the sea wave video, the ultra-wide angle sky video and the meteorological observation data to obtain a copied sea wave video, a copied sky video and copied meteorological data; copying the wave videos comprises copying a first long-focus wave video, copying a second long-focus wave video, copying a third long-focus wave video and copying a wide-angle wave video;
establishing a plurality of folders according to the data acquisition time and the video types;
respectively storing a first copy of the long-focus sea wave video, a second copy of the long-focus sea wave video, a third copy of the long-focus sea wave video, a copy of the wide-angle sea wave video, a copy of the sky video and a copy of weather data into a plurality of folders;
determining a first copy of the long-focus sea wave video, a second copy of the long-focus sea wave video, a third copy of the long-focus sea wave video and a copy of the wide-angle sea wave video in the plurality of folders as packaged sea wave data, determining a copy sky video as packaged sky data, and determining a copy weather data as packaged weather data.
In this embodiment, the method can perform secondary encapsulation on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data through the computer. Specifically, the method can classify the video images and the data related to the automatic weather station into folders according to two categories. The video images can be named and stored correspondingly according to the data acquisition time, for example, a tele-camera A1 is placed in a folder A1; and storing the meteorological observation data in a meteorological observation data folder.
In this embodiment, because the ocean observation data is very important, the data to be processed needs to be copied, so as to avoid the change of the original data. In this way, when the subsequent processing program is in error, the folder data needing to be read is lost, and the previous data can be acquired from the original data folder.
S103, screening the packaged sea wave data according to the packaged sky data and the packaged meteorological data to obtain effective sea wave data.
As an optional implementation manner, the step of performing screening processing on the encapsulated wave data according to the encapsulated sky data and the encapsulated weather data to obtain effective wave data includes:
determining an effective acquisition time period according to the encapsulated sky data; the effective acquisition time period is the time period from sunrise to sunset;
extracting daytime sea wave data from the packaged sea wave data according to an effective acquisition time period;
determining an invalid acquisition time period according to the packaged meteorological data;
and extracting effective sea wave data from the sea wave data in the daytime according to the ineffective acquisition time period.
In this embodiment, because wave and weather conditions can not be effectively collected at night, data of the time period from sunrise to sunset need to be acquired. This is indicative of automated acquisition.
In this embodiment, the local ocean time zone versus the sunrise time and the sunset time can be determined according to the wide-angle sky data, so as to select the time period of the day, for example, six am is the sunrise time, six pm is the sunset time, and the data from six am and half to five pm is preferentially selected as the daytime data.
In this embodiment, after the wave data in the daytime is acquired, the quality of the acquired data is poor due to problems such as weather phenomena and visibility. For example, when the weather phenomenon is rain or the visibility is lower than 10 kilometers, the video images of the type are poor in quality and need to be removed.
In this embodiment, the basic condition of the culling is that the video image data is not clear enough, because the video image that is not clear enough directly affects the efficiency and quality of the subsequent deep learning. Therefore, in rainy days and foggy days, for example, the condition that the weather condition with stronger shielding is not favorable for shooting triggers a rejection condition, so that the data in the invalid time periods are rejected to ensure that the shot pictures are clear. The reason why the conditions such as rainy days and foggy days are selected is that the weather condition can be automatically judged by the weather visibility meter, and the visibility can be recorded at the same time. Whether the weather condition accords with the shooting condition or not can be known through the combined judgment of weather and visibility, so that the data are judged to be effective automatically, and the effect of automatically acquiring the data is realized. After all, the whole shooting period is long, a lot of video images are needed, and only then is enough samples available for deep learning training, so that the automatic judgment is the basis of subsequent deep learning training.
S104, performing deep learning modeling on effective sea wave data according to a preset GPU (graphics processing unit) convolutional neural network to obtain a first wave-level observation model; the effective sea wave data is provided with a training label of the sea wave grade.
As an optional implementation manner, the step of performing deep learning modeling on the effective sea wave data according to a preset GPU convolutional neural network to obtain a first wave level observation model includes:
extracting a plurality of image pictures in effective sea wave data;
splitting each image picture into 9 image sub-pictures according to the resolution of 640 x 360;
dividing all image sprites into a training set and a verification set;
and performing deep learning modeling according to the training set, the verification set and a preset GPU convolutional neural network to obtain a first wave-level observation model.
In this embodiment, compared with CPU training, GPU training has a higher calculation speed, and the total power consumption is lower than that of the CPU, which is generally high-speed and low-consumption, not only facilitates work progress, but also achieves energy saving and emission reduction effects for carbon neutralization.
In this embodiment, the GPU convolutional neural network preferentially selects the fully-connected neural network, and processes the screened effective ocean wave data. Specifically, the method first splits the effective ocean wave data, for example, splits a picture into 3 × 3 small pictures, for example, into 640 × 360 resolutions. Therefore, the result that a picture at one moment generates a plurality of groups of effective observation data can be realized.
In this embodiment, the number ratio between the training set and the validation set is 8: 2.
In this embodiment, it should be noted that the effective wave data is usually a wave video image with a relatively high wave level.
S105, judging whether the model precision of the first wave-level observation model is higher than the preset precision, if so, executing the step S106; if not, improving the effective sea wave data, and triggering and executing the step S104.
In this embodiment, the method preferentially adopts effective wave data for training, and after the accuracy of the first-wave-level observation model is not lower than 80%, the element wind is added as required (i.e., step S106 is executed).
In this embodiment, the observation accuracy is fixed. It is understood that the accuracy (resolution) of the camera shooting is fixed, so that the observation accuracy is fixed.
In this embodiment, the model precision is used to represent the observation accuracy that the wave-level observation model can achieve. For example, the model precision is used to represent the accuracy of the output result obtained by the wave-level observation model after receiving the data input.
In the present embodiment, the model accuracy is generally affected by factors such as image resolution and data size. This makes the test integration one of the factors that have a large impact on the training effect of the model in practice (i.e. the test set accuracy is the relatively closest judgment standard to the practice), so the method can improve the model accuracy by improving the training set accuracy.
It can be understood that the method can use the test set to test the wave-level observation model, and the precision of the test set is obtained. The test set accuracy is the model accuracy described above.
In this embodiment, the purpose of improving the valid data is to improve the validity of the training set. The improvement method may include a method of expanding the volume of the data, a method of refining the data, and the like.
For example, the method may increase the model accuracy of the first wave-level observation model by adding the training set, so that the model accuracy of the first wave-level observation model reaches the final preset accuracy.
In this embodiment, the subsequent training process may also be optimized and improved in this way.
In this embodiment, the training of the first-wave observation model needs to be ensured to be completed in the process that the method needs to be implemented. In the step, a cycle improvement method is used, so that the first wave-level observation model is practically trained. Thereby providing a basis for the subsequent deep learning process. In short, when the problem of invalid data part is encountered, the improvement can be carried out by a method of eliminating invalid data; when the problem that the model precision is not high is met, the improvement can be carried out by adding effective data. In conclusion, the method can ensure that the training of the first wave-level observation model is completed.
And S106, carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
In this embodiment, since the effective sea wave data exists in the form of pixels, the wind speed data in the meteorological observation field data should also be in the form of images. For example, 3 meters of wind, the picture is formed into a 640 x 360 resolution picture corresponding to the picture in the form of a pure digital matrix 03, 03, 03. And finally, when data reading is carried out, reading the processed data into an initial field according to different channels for training.
As an optional implementation, the deep learning modeling is performed according to the meteorological observation data and the first wave-level observation model, and the step of obtaining the ocean wave-level observation model includes:
extracting meteorological observation data including wind speed data, wind direction data, temperature data, air pressure data, humidity data and visibility data;
performing deep learning modeling according to the wind speed data and the first wave-level observation model to obtain a second wave-level observation model;
judging whether the model precision of the second wave-level observation model is higher than the preset precision or not;
when the model precision of the second wave-level observation model is higher than the preset precision, performing deep learning modeling according to the wind direction data and the second wave-level observation model to obtain a third wave-level observation model;
judging whether the model precision of the third wave-level observation model is higher than the preset precision or not;
when the model precision of the third wave-level observation model is higher than the preset precision, performing deep learning modeling according to the temperature data and the third wave-level observation model to obtain a fourth wave-level observation model;
judging whether the model precision of the fourth wave-level observation model is higher than the preset precision or not;
when the model precision of the fourth wave-level observation model is higher than the preset precision, performing deep learning modeling according to the air pressure data and the fourth wave-level observation model to obtain a fifth wave-level observation model;
judging whether the model precision of the fifth wave-level observation model is higher than the preset precision or not;
when the model precision of the fifth wave-level observation model is higher than the preset precision, performing deep learning modeling according to the humidity data and the fifth wave-level observation model to obtain a sixth wave-level observation model;
judging whether the model precision of the sixth wave-level observation model is higher than the preset precision or not;
when the model precision of the sixth wave-level observation model is higher than the preset precision, performing deep learning modeling according to the visibility data and the sixth wave-level observation model to obtain an ocean wave-level observation model;
when the model precision of the second wave-level observation model, the model precision of the third wave-level observation model, the model precision of the fourth wave-level observation model, the model precision of the fifth wave-level observation model or the model precision of the sixth wave-level observation model is not higher than the preset precision, the second wave-level observation model, the third wave-level observation model, the fourth wave-level observation model, the fifth wave-level observation model or the sixth wave-level observation model is correspondingly determined as the ocean wave-level observation model.
In this embodiment, because wave level training in the ocean is not easy, if the relevant elements are aligned at one time, the acquisition of the training result becomes very slow, and the accuracy is affected to some extent by black box training. Therefore, the method can realize a stable modeling effect of firstly preserving the bottom and then adding the quantity.
As an optional implementation, the deep learning modeling is performed according to the meteorological observation data and the first wave-level observation model, and the step of obtaining the ocean wave-level observation model includes:
extracting meteorological observation data including wind speed data, wind direction data, temperature data, air pressure data, humidity data and visibility data;
performing deep learning modeling according to the wind speed data and the first wave-level observation model to obtain a second wave-level observation model;
performing deep learning modeling according to the wind direction data and the first wave-level observation model to obtain a third wave-level observation model;
performing deep learning modeling according to the temperature data and the first wave-level observation model to obtain a fourth wave-level observation model;
performing deep learning modeling according to the air pressure data and the first wave-level observation model to obtain a fifth wave-level observation model;
performing deep learning modeling according to the humidity data and the first wave-level observation model to obtain a sixth wave-level observation model;
and carrying out corresponding model fusion on the second wave-level observation model, the third wave-level observation model, the fourth wave-level observation model, the fifth wave-level observation model and the sixth wave-level observation model according to the model precision of the second wave-level observation model, the third wave-level observation model, the fourth wave-level observation model and the sixth wave-level observation model to obtain the oceanic wave-level observation model.
As an optional implementation, after performing deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain the ocean wave-level observation model, the method further includes:
acquiring a current sea wave video image;
and inputting the sea wave video image into the ocean wave level observation model to obtain the sea wave level.
In this embodiment, the method may acquire a large amount of data (the large amount of data includes data of a tele camera a1, data of a tele camera a4, data of a wide camera a7, data of a tele camera a10, data of an ultra wide camera B1, and packaged sky data including true wind speed, true wind direction, air pressure, temperature, humidity, visibility, and weather phenomena, and meanwhile, the large amount of data may further include a GPS coordinate position, a ship heading, a ship speed, and a sea level obtained through identification) by a ship-mounted device, and send the large amount of data to the cloud platform.
In this embodiment, the cloud platform may display the large amount of data in different colors (the colors may be labeled with reference to the colors of the sea wave disaster level in our country). If the cloud platform identifies the blue sea wave grade level, further acquiring a shot video acquired by the shipborne equipment, wherein the shot video does not exceed ten seconds generally.
In this embodiment, the cloud platform may further provide query modes based on time, longitude and latitude, region, and the like. Besides the operators, sea wave grade modification permission is provided for authenticated national-grade ocean forecast service personnel to correct the wrong position of deep-learning sea wave grade judgment, the modification and the deep-learning sea wave grade model judgment grade are synchronously displayed, and if the manual modification is correct, data support is provided for a subsequent optimized ocean wave grade observation model.
In the embodiment, a large amount of effective data can be acquired through the application of the cloud platform, so that the ocean wave level observation model can be further optimized, and the accuracy of ocean wave level observation in the ocean is improved to the maximum extent.
By implementing the embodiment, the real sea state of the open sea (ocean) can be effectively observed and recorded, the sea state cannot be observed in the offshore region, so that the waves and swell in the waves cannot be effectively observed only by the waves near the shore in the offshore observation, which is a specific advantage of the method. Meanwhile, the method can be used for carrying out wave grade division based on the completed deep learning model (ocean wave level observation model), so that ocean waves can be recognized with high precision. In addition, the method can share the photos received by the ocean wave level observation model, the videos and the ocean wave levels output by the ocean wave level observation model through the cloud platform, so that the ocean sea state can be shared, and corresponding early warning can be conveniently carried out on the ocean sea state based on the shared data; on the basis of the cloud platform, the cloud platform also allows professionals to modify the sea wave grade manually, so that the final sea wave grade is more objective and fair.
In this embodiment, the execution subject of the method may be a computing system such as a computer and a server, and is not limited in this embodiment.
In this embodiment, the meanings of the method and the similar descriptions in the open sea and the deep sea are all the meanings of the ocean, and the description thereof is omitted.
It can be seen that, by implementing the ocean wave observation model training method described in this embodiment, a neural network model for observing ocean wave levels in an ocean scene can be trained, so that when a corresponding device acquires an ocean image, ocean waves can be automatically identified to obtain ocean wave levels. The application of the model can realize the observation of low cost, high precision and high stability on ocean wave grades, thereby providing powerful data support for commercial ships, fishermen, ocean scientific research and business personnel and the like.
Example 2
Referring to fig. 2, fig. 2 is a schematic structural diagram of an ocean wave observation model training system according to this embodiment. As shown in fig. 2, the training system for ocean wave observation model includes:
the acquiring unit 210 is configured to acquire a sea wave video, an ultra-wide-angle sky video, and meteorological observation data; the wave video comprises a first long-focus wave video, a second long-focus wave video, a third long-focus wave video and a wide-angle wave video;
the packaging unit 220 is used for performing secondary packaging on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain packaged sea wave data, packaged sky data and packaged meteorological data;
the screening unit 230 is configured to perform screening processing on the packaged wave data according to the packaged sky data and the packaged meteorological data to obtain effective wave data;
the modeling unit 240 is used for performing deep learning modeling on effective sea wave data according to a preset GPU convolutional neural network to obtain a first wave-level observation model; training labels of sea wave grades are set in the effective sea wave data;
and the modeling unit 240 is further configured to perform deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
As an alternative embodiment, the encapsulation unit 220 includes:
the copying subunit 221 is configured to copy the sea wave video, the ultra-wide-angle sky video and the weather observation data to obtain a copied sea wave video, a copied sky video and copied weather data; copying the wave videos comprises copying a first long-focus wave video, copying a second long-focus wave video, copying a third long-focus wave video and copying a wide-angle wave video;
a creating subunit 222 configured to create a plurality of folders according to the data acquisition time and the video genre;
a storage subunit 223, configured to store the first copy long-focus sea wave video, the second copy long-focus sea wave video, the third copy long-focus sea wave video, the copy wide-angle sea wave video, the copy sky video, and the copy weather data into a plurality of folders, respectively;
and the packaging subunit 224 is configured to determine the first copied long-focus sea wave video, the second copied long-focus sea wave video, the third copied long-focus sea wave video and the copied wide-angle sea wave video in the plurality of folders as packaged sea wave data, determine the copied sky video as packaged sky data, and determine the copied weather data as packaged weather data.
As an alternative embodiment, the screening unit 230 includes:
a determining subunit 231, configured to determine an effective acquisition time period according to the encapsulated sky data; the effective acquisition time period is the time period from sunrise to sunset;
a screening subunit 232, configured to extract daytime sea wave data from the encapsulated sea wave data according to the effective acquisition time period;
the determining subunit 231 is further configured to determine an invalid acquisition time period according to the packaged meteorological data;
the screening subunit 232 is further configured to extract valid sea wave data from the sea wave data in the daytime according to the invalid acquisition time period.
As an alternative embodiment, the modeling unit 240 includes:
an extracting subunit 241, configured to extract a plurality of image frames in the effective ocean wave data;
a splitting subunit 242, configured to split each image picture into 9 image sub-pictures according to a resolution of 640 × 360;
a dividing subunit 243, configured to divide all image sprites into a training set and a verification set;
and a modeling unit 244 for performing deep learning modeling according to the training set, the verification set and a preset GPU convolutional neural network to obtain a first wave-level observation model.
As an optional implementation, the training system for the ocean wave observation model further includes:
a determining unit 250, configured to determine whether a model precision of the first wave-level observation model is higher than a preset precision;
and the modeling unit 240 is specifically configured to, when the model precision of the first wave-level observation model is higher than the preset precision, perform deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an operation of the ocean wave-level observation model.
As an alternative embodiment, the modeling unit 240 includes:
the extracting subunit 241 is configured to extract meteorological observation data including wind speed data, wind direction data, temperature data, air pressure data, humidity data, and visibility data;
the modeling unit 244 is used for performing deep learning modeling according to the wind speed data and the first wave-level observation model to obtain a second wave-level observation model;
a determining subunit 245, configured to determine whether the model precision of the second wave-level observation model is higher than a preset precision;
the modeling subunit 244 is further configured to perform deep learning modeling on the wind direction data and the second wave-level observation model to obtain a third wave-level observation model;
the determining subunit 245 is further configured to determine whether the model precision of the third wave-level observation model is higher than the preset precision;
the modeling subunit 244 is further configured to, when the model accuracy of the third wave-level observation model is higher than the preset accuracy, perform deep learning modeling according to the temperature data and the third wave-level observation model to obtain a fourth wave-level observation model;
the determining subunit 245 is further configured to determine whether the model precision of the fourth wave-level observation model is higher than the preset precision;
the modeling subunit 244 is further configured to, when the model precision of the fourth wave-level observation model is higher than the preset precision, perform deep learning modeling according to the air pressure data and the fourth wave-level observation model to obtain a fifth wave-level observation model;
the determining subunit 245 is further configured to determine whether the model precision of the fifth wave-level observation model is higher than the preset precision;
the modeling subunit 244 is further configured to, when the model precision of the fifth wave-level observation model is higher than the preset precision, perform deep learning modeling according to the humidity data and the fifth wave-level observation model to obtain a sixth wave-level observation model;
the determining subunit 245 is further configured to determine whether the model precision of the sixth wave-level observation model is higher than the preset precision;
the modeling subunit 244 is further configured to, when the model precision of the sixth-wave-level observation model is higher than a preset precision, perform deep learning modeling according to the visibility data and the sixth-wave-level observation model to obtain an ocean-wave-level observation model;
the modeling subunit 244 is further configured to, when the model precision of the second wave-level observation model, the model precision of the third wave-level observation model, the model precision of the fourth wave-level observation model, the model precision of the fifth wave-level observation model, or the model precision of the sixth wave-level observation model is not higher than the preset precision, correspondingly determine the second wave-level observation model, the third wave-level observation model, the fourth wave-level observation model, the fifth wave-level observation model, or the sixth wave-level observation model as the ocean wave-level observation model.
As an alternative embodiment, the modeling unit 240 includes:
the extracting subunit 241 is configured to extract meteorological observation data including wind speed data, wind direction data, temperature data, air pressure data, humidity data, and visibility data;
the modeling unit 244 is used for performing deep learning modeling according to the wind speed data and the first wave-level observation model to obtain a second wave-level observation model;
the modeling subunit 244 is further configured to perform deep learning modeling according to the wind direction data and the first wave-level observation model to obtain a third wave-level observation model;
the modeling subunit 244 is further configured to perform deep learning modeling according to the temperature data and the first wave-level observation model to obtain a fourth wave-level observation model;
the modeling subunit 244 is further configured to perform deep learning modeling according to the air pressure data and the first wave-level observation model to obtain a fifth wave-level observation model;
the modeling subunit 244 is further configured to perform deep learning modeling according to the humidity data and the first wave-level observation model to obtain a sixth wave-level observation model;
the modeling subunit 244 is further configured to perform corresponding model fusion on the second wave level observation model, the third wave level observation model, the fourth wave level observation model, the fifth wave level observation model, and the sixth wave level observation model according to the model accuracy of the second wave level observation model, the model accuracy of the third wave level observation model, the model accuracy of the fourth wave level observation model, the model accuracy of the fifth wave level observation model, and the model accuracy of the sixth wave level observation model, so as to obtain the ocean wave level observation model.
In the embodiment of the present application, for the explanation of the training system for the ocean wave observation model, reference may be made to the description in embodiment 1, and details are not repeated in this embodiment.
It can be seen that, the ocean wave observation model training system described in this embodiment can train a neural network model for observing ocean wave levels in an ocean scene, so that corresponding devices can automatically identify ocean waves to obtain ocean wave levels when acquiring ocean images. The application of the model can realize the observation of low cost, high precision and high stability on ocean wave grades, thereby providing powerful data support for commercial ships, fishermen, ocean scientific research and business personnel and the like.
The embodiment of the present application provides an electronic device, which includes a memory and a processor, where the memory is used to store a computer program, and the processor runs the computer program to make the electronic device execute the ocean wave observation model training method in embodiment 1 of the present application.
The embodiment of the present application provides a computer-readable storage medium, which stores computer program instructions, and when the computer program instructions are read and executed by a processor, the method for training the ocean wave observation model in embodiment 1 of the present application is performed.
In the several embodiments provided in the present application, it should be understood that the disclosed system and method may be implemented in other ways. The above-described system embodiments are merely illustrative, and for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. An ocean wave observation model training method is characterized by comprising the following steps:
obtaining sea wave video, ultra-wide angle sky video and meteorological observation data; the wave video comprises a first long-focus wave video, a second long-focus wave video, a third long-focus wave video and a wide-angle wave video;
performing secondary packaging on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain packaged sea wave data, packaged sky data and packaged meteorological data;
screening the packaged sea wave data according to the packaged sky data and the packaged meteorological data to obtain effective sea wave data;
according to a preset GPU convolutional neural network, performing deep learning modeling on the effective sea wave data to obtain a first wave-level observation model; training labels of sea wave grades are set in the effective sea wave data;
and carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
2. The ocean wave-level observation model training method of claim 1, wherein the step of secondarily encapsulating the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain encapsulated sea wave data, encapsulated sky data and encapsulated meteorological data comprises:
copying the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain a copied sea wave video, a copied sky video and copied meteorological data; the copying wave video comprises a first copying long-focus wave video, a second copying long-focus wave video, a third copying long-focus wave video and a copying wide-angle wave video;
establishing a plurality of folders according to the data acquisition time and the video types;
storing the first copy of the tele sea wave video, the second copy of the tele sea wave video, the third copy of the tele sea wave video, the copy of the wide-angle sea wave video, the copy of the sky video and the copy of the weather data into the plurality of folders respectively;
determining the first copy of the tele sea wave video, the second copy of the tele sea wave video, the third copy of the tele sea wave video and the copy of the wide sea wave video in the plurality of folders as packaged sea wave data, determining the copy sky video as packaged sky data, and determining the copy weather data as packaged weather data.
3. The ocean wave observation model training method according to claim 1, wherein the step of screening the encapsulated ocean wave data according to the encapsulated sky data and the encapsulated weather data to obtain effective ocean wave data comprises:
determining an effective acquisition time period according to the encapsulated sky data; the effective acquisition time period is a time period from sunrise to sunset;
extracting daytime sea wave data from the encapsulated sea wave data according to the effective acquisition time period;
determining an invalid acquisition time period according to the packaged meteorological data;
and extracting effective sea wave data from the daytime sea wave data according to the ineffective acquisition time interval.
4. The training method of the ocean wave-level observation model according to claim 1, wherein the step of performing deep learning modeling on the effective ocean wave data according to a preset GPU convolutional neural network to obtain a first wave-level observation model comprises:
extracting a plurality of image pictures in the effective sea wave data;
splitting each image picture into 9 image sub-pictures according to the resolution of 640 x 360;
dividing all image sprites into a training set and a verification set;
and performing deep learning modeling according to the training set, the verification set and a preset GPU convolutional neural network to obtain a first wave-level observation model.
5. The ocean wave scale observation model training method of claim 1, further comprising:
judging whether the model precision of the first wave-level observation model is higher than a preset precision or not;
and when the model precision of the first wave-level observation model is higher than the preset precision, executing the step of carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
6. The method for training the ocean wave-level observation model according to claim 1, wherein the step of performing deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain the ocean wave-level observation model comprises:
extracting the meteorological observation data comprising wind speed data, wind direction data, temperature data, air pressure data, humidity data and visibility data;
performing deep learning modeling according to the wind speed data and the first wave-level observation model to obtain a second wave-level observation model;
judging whether the model precision of the second wave-level observation model is higher than a preset precision or not;
when the model precision of the second wave-level observation model is higher than the preset precision, performing deep learning modeling according to the wind direction data and the second wave-level observation model to obtain a third wave-level observation model;
judging whether the model precision of the third wave-level observation model is higher than the preset precision or not;
when the model precision of the third wave-level observation model is higher than the preset precision, performing deep learning modeling according to the temperature data and the third wave-level observation model to obtain a fourth wave-level observation model;
judging whether the model precision of the fourth wave-level observation model is higher than the preset precision or not;
when the model precision of the fourth wave-level observation model is higher than the preset precision, performing deep learning modeling according to the air pressure data and the fourth wave-level observation model to obtain a fifth wave-level observation model;
judging whether the model precision of the fifth wave-level observation model is higher than the preset precision or not;
when the model precision of the fifth wave-level observation model is higher than the preset precision, performing deep learning modeling according to the humidity data and the fifth wave-level observation model to obtain a sixth wave-level observation model;
judging whether the model precision of the sixth wave-level observation model is higher than the preset precision or not;
when the model precision of the sixth wave-level observation model is higher than the preset precision, performing deep learning modeling according to the visibility data and the sixth wave-level observation model to obtain an ocean wave-level observation model;
when the model precision of the second wave level observation model, the model precision of the third wave level observation model, the model precision of the fourth wave level observation model, the model precision of the fifth wave level observation model or the model precision of the sixth wave level observation model is not higher than the preset precision, the second wave level observation model, the third wave level observation model, the fourth wave level observation model, the fifth wave level observation model or the sixth wave level observation model is correspondingly determined as the ocean wave level observation model.
7. The method for training the ocean wave-level observation model according to claim 1, wherein the step of performing deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain the ocean wave-level observation model comprises:
extracting the meteorological observation data comprising wind speed data, wind direction data, temperature data, air pressure data, humidity data and visibility data;
performing deep learning modeling according to the wind speed data and the first wave-level observation model to obtain a second wave-level observation model;
performing deep learning modeling according to the wind direction data and the first wave-level observation model to obtain a third wave-level observation model;
performing deep learning modeling according to the temperature data and the first wave-level observation model to obtain a fourth wave-level observation model;
performing deep learning modeling according to the air pressure data and the first wave-level observation model to obtain a fifth wave-level observation model;
performing deep learning modeling according to the humidity data and the first wave-level observation model to obtain a sixth wave-level observation model;
and carrying out corresponding model fusion on the second wave level observation model, the third wave level observation model, the fourth wave level observation model, the fifth wave level observation model and the sixth wave level observation model according to the model precision of the second wave level observation model, the model precision of the third wave level observation model, the model precision of the fourth wave level observation model, the model precision of the fifth wave level observation model and the model precision of the sixth wave level observation model to obtain the ocean wave level observation model.
8. An ocean wave scale observation model training system, comprising:
the acquisition unit is used for acquiring sea wave videos, ultra-wide-angle sky videos and meteorological observation data; the wave video comprises a first long-focus wave video, a second long-focus wave video, a third long-focus wave video and a wide-angle wave video;
the packaging unit is used for carrying out secondary packaging on the sea wave video, the ultra-wide-angle sky video and the meteorological observation data to obtain packaged sea wave data, packaged sky data and packaged meteorological data;
the screening unit is used for screening the packaged sea wave data according to the packaged sky data and the packaged meteorological data to obtain effective sea wave data;
the modeling unit is used for carrying out deep learning modeling on the effective sea wave data according to a preset GPU convolutional neural network to obtain a first wave-level observation model; training labels of sea wave grades are set in the effective sea wave data;
the modeling unit is further used for carrying out deep learning modeling according to the meteorological observation data and the first wave-level observation model to obtain an ocean wave-level observation model.
9. An electronic device, comprising a memory for storing a computer program and a processor for executing the computer program to cause the electronic device to perform the ocean wave observation model training method of any one of claims 1 to 7.
10. A readable storage medium having stored thereon computer program instructions which, when read and executed by a processor, perform the ocean wave observation model training method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210112584.6A CN114519806B (en) | 2022-01-29 | 2022-01-29 | Ocean wave level observation model training method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210112584.6A CN114519806B (en) | 2022-01-29 | 2022-01-29 | Ocean wave level observation model training method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114519806A true CN114519806A (en) | 2022-05-20 |
| CN114519806B CN114519806B (en) | 2022-10-11 |
Family
ID=81597458
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210112584.6A Active CN114519806B (en) | 2022-01-29 | 2022-01-29 | Ocean wave level observation model training method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114519806B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109359787A (en) * | 2018-12-06 | 2019-02-19 | 上海海事大学 | A kind of multi-modal wave forecasting system in small range sea area and its prediction technique |
| CN109886217A (en) * | 2019-02-26 | 2019-06-14 | 上海海洋大学 | A method of it is high that wave being detected from Nearshore Wave video based on convolutional neural networks |
| CN110232342A (en) * | 2019-06-03 | 2019-09-13 | 中国人民解放军海军航空大学 | Sea situation level determination method and device based on convolutional neural networks |
| CN111736148A (en) * | 2020-06-28 | 2020-10-02 | 国家海洋环境预报中心 | Method for correcting sea wave effective wave height of satellite radar altimeter and related device |
| CN113033094A (en) * | 2021-03-24 | 2021-06-25 | 上海海洋大学 | Sea wave height prediction method |
| CN113221777A (en) * | 2021-05-19 | 2021-08-06 | 江苏奥易克斯汽车电子科技股份有限公司 | Offshore sea wave grade monitoring method and device |
| US20220003894A1 (en) * | 2018-09-26 | 2022-01-06 | Sofar Ocean Technologies, Inc. | Ocean weather forecasting system |
| CN113899349A (en) * | 2021-10-26 | 2022-01-07 | 湖北中南鹏力海洋探测系统工程有限公司 | Sea wave parameter detection method, equipment and storage medium |
-
2022
- 2022-01-29 CN CN202210112584.6A patent/CN114519806B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220003894A1 (en) * | 2018-09-26 | 2022-01-06 | Sofar Ocean Technologies, Inc. | Ocean weather forecasting system |
| CN109359787A (en) * | 2018-12-06 | 2019-02-19 | 上海海事大学 | A kind of multi-modal wave forecasting system in small range sea area and its prediction technique |
| CN109886217A (en) * | 2019-02-26 | 2019-06-14 | 上海海洋大学 | A method of it is high that wave being detected from Nearshore Wave video based on convolutional neural networks |
| CN110232342A (en) * | 2019-06-03 | 2019-09-13 | 中国人民解放军海军航空大学 | Sea situation level determination method and device based on convolutional neural networks |
| CN111736148A (en) * | 2020-06-28 | 2020-10-02 | 国家海洋环境预报中心 | Method for correcting sea wave effective wave height of satellite radar altimeter and related device |
| CN113033094A (en) * | 2021-03-24 | 2021-06-25 | 上海海洋大学 | Sea wave height prediction method |
| CN113221777A (en) * | 2021-05-19 | 2021-08-06 | 江苏奥易克斯汽车电子科技股份有限公司 | Offshore sea wave grade monitoring method and device |
| CN113899349A (en) * | 2021-10-26 | 2022-01-07 | 湖北中南鹏力海洋探测系统工程有限公司 | Sea wave parameter detection method, equipment and storage medium |
Non-Patent Citations (5)
| Title |
|---|
| HEEJEONG CHOI 等: "Real-time significant wave height estimation from raw ocean images based on 2D and 3D deep neural networks", 《ELSEVIER》 * |
| WEI SONG 等: "Determining Wave Height from Nearshore Videos Based on Multi-level Spatiotemporal Feature Fusion", 《2021 IEEE》 * |
| 孟雷: "神经网络方法对海浪有效波高数值模拟的改进", 《海洋预报》 * |
| 林民龙: "基于神经网络集成的增量式学习", 《中国优秀硕士学位论文全文数据库 工程科技I辑》 * |
| 郝剑波: "基于深度学习的近岸海浪等级分类研究", 《中国优秀硕士学位论文全文数据库 基础科学辑》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114519806B (en) | 2022-10-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hess et al. | Geocoded digital videography for validation of land cover mapping in the Amazon basin | |
| CN113076842B (en) | Method for improving traffic sign recognition accuracy in extreme weather and environment | |
| CN110110641B (en) | Unmanned aerial vehicle monitoring method and system for watershed flooding scene | |
| Meaden et al. | Geographical information systems: applications to marine fisheries | |
| JP5283214B2 (en) | Fixed point observation apparatus and fixed point observation method | |
| CN112488020B (en) | Water environment pollution condition detection and evaluation device based on unmanned aerial vehicle aerial photographing data | |
| CN103605171A (en) | All-sky imaging instrument and cloud layer characteristic analysis method based on all-sky imaging instrument | |
| CN111145136A (en) | Synthesis method, system and storage medium for transformer substation meter image data set | |
| Chen et al. | Real-time garbage object detection with data augmentation and feature fusion using SUAV low-altitude remote sensing images | |
| CN116223511A (en) | Distributed roof photovoltaic module defect diagnosis method and device based on unmanned aerial vehicle automatic inspection | |
| CN114519806B (en) | Ocean wave level observation model training method and system | |
| He et al. | A novel image recognition algorithm of target identification for unmanned surface vehicles based on deep learning | |
| CN112581301A (en) | Detection and early warning method and system for residual film quantity of farmland based on deep learning | |
| CN112785678B (en) | Sunlight analysis method and system based on three-dimensional simulation | |
| Petković et al. | Image Dataset for Neural Network Performance Estimation with Application to Maritime Ports | |
| Xu et al. | UAV-ODS: A real-time outfall detection system based on UAV remote sensing and edge computing | |
| CN115760885B (en) | High-closure-degree wetland forest parameter extraction method based on consumer-level unmanned aerial vehicle image | |
| CN116109942A (en) | Ship target detection method for visible light remote sensing image | |
| CN113867404B (en) | Beach garbage inspection method and system based on unmanned aerial vehicle | |
| Long et al. | Whiting events in SW Florida coastal waters: A case study using MODIS medium-resolution data | |
| Sullivan et al. | Electric transmission visibility and visual contrast threshold distances in western landscapes | |
| Amery | Capturing the Northern Lights: Standardizing the practice of auroral photography during the Second International Polar Year, 1932–1933 | |
| Pfeiffer et al. | Detecting beach litter in drone images using deep learning | |
| CN113505699A (en) | Ship detection method based on RetinaNet algorithm | |
| CN108052585B (en) | Method for judging dynamic target in complex environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |